Systemic risk forum

We held the long awaited Forum on Systemic Stability and Liquidity this week at Fields. This was one of the activities that I was looking forward to the most in the thematic program, not only because it relates directly to the financial crises, but because it brought to Fields a diverse group of speakers, including some renowned economists, who are hard to find in typical mathematics conference.

Wei Xiong opened the proceedings with a talk incorporating financing choices and diverse beliefs into asset pricing and associated booms and crashes. This naturally delighted me because of my recent obsession with finding mathematical formulations for the Minsky story about asset bubbles and fragility of financial systems.

Jennifer Huang followed with a simple model explaining market failures in liquidity provision which allowed her to analyze the effects of several different policy interventions, most of them used in the aftermath of the crisis of 2008.

Rama Cont concluded the morning with his network model for banking systems and the corresponding measures for the default impact of any single bank and the associated systemic risk that it poses on the network. I was particularly interested in this talk, since one of my PhD students, Omneia Ismail, is currently working on an agent-based extension of this model.

Hao Zhou, the first of three speakers from the Fed, presented an intriguing way of measuring systemic risk by constructing a hypothetical insurance contract covering the losses of the entire banking sector. Although such contract would never be sold by any individual insurance company (for obvious reasons), the marginal contribution of each bank to the hypothetical premium is an indication of the risk it poses to the system.

In a different direction, Kay Giesecke proposed to link systemic risk directly with the probability of default of a large fraction of the financial sector and suggested a clever two-step maximum likelihood method to estimate this probability from default events occurring in the entire economy.

The first day ended with a panel discussion featuring Aaron Brown, Michael Gordy, Joseph Langsam and David Rowe and provided unique insights on key factors leading up to the crisis and a lively discussion on policy, legislation and general banking practices.

Frank Milne started the second day with a tour de force on roughly sixty years of attempts to introduce illiquidity into traditional economic models (read general equilibrium a la Arrow-Debreu). He knows all the tricks in this trade, and anyone working in this area should bounce their ideas off him to make sure that they are neither reinventing the wheel or moving towards a dead end. What got him really excited towards the end of his talk was the idea of financial war games, which sits squarely within the agent-based models for financial systems that Omneia and I are working on.

Viral Acharya proposed yet any way to measure systemic risk based on the degree of under-capitalization of a financial institute. This leads to the concept of Marginal Expected Shortfall, which I still think should have the words "systemic" in it, just so that they can properly call it a MESS (this was my piece of advice to Robert Engle when I heard him talk about it in the econometrics workshop last month) 

Itay Goldstein presented a game-theoretical model for credit freezes, where the lending actions of banks reinforce the probability of successful outcomes for the loans of other banks. He showed how this can lead to different equilibria: efficient no-lending (when the economic outlook is sufficiently bad that no bank would recover their loans, efficient lending (when the economic outlook is so good that all loans are fully recovered no matter what) and an interesting regime where both lending and no-lending are possible equilibria, and credit can freeze due to lack of coordination between banks. He then analyze the effect of different policy interventions according to this model.

The last two talks of the forum were dedicated to the concept of risk appetite. Jon Danielsson provided the theoretical background by explaining that risk appetite is the result of risk preferences and beliefs under constraints, and therefore can very wildly in the market as constraints move from non-biding to biding and back again. Erkko Etula provided empirical evidence for exactly this type of phenomenon using constraints in the funding liquidity for US financial intermediaries. The associated changes in the corresponding risk appetite can then be used to forecast changes in exchange rate between the US dollars and most major currencies.